The only thing more amusing than the weekly upward revisions in the initial weekly jobless claims are the media reports of significant drops afterward.  Today’s case in point is the “drop” to 377,000 from last week’s 383,000, which got revised upward to 389,000 — the 64th upward revision in 65 weeks:

In the week ending June 2, the advance figure for seasonally adjusted initial claims was 377,000, a decrease of 12,000 from the previous week’s revised figure of 389,000. The 4-week moving average was 377,750, an increase of 1,750 from the previous week’s revised average of 376,000.

The advance seasonally adjusted insured unemployment rate was 2.6 percent for the week ending May 26, unchanged from the prior week’s unrevised rate.

The advance number for seasonally adjusted insured unemployment during the week ending May 26 was 3,293,000, an increase of 34,000 from the preceding week’s revised level of 3,259,000. The 4-week moving average was 3,279,500, an increase of 11,500 from the preceding week’s revised average of 3,268,000.

First, let’s make the extent of the increases and declines clear.  A drop from 383K to 377K would have been just slightly over 1%.  Going from 389K to 377 is a 3% change.  These are basically variations within the same statistical band.  It’s true that 377K is better than 389K, but it’s hardly a big shift in momentum — even if we could rely on the number, and not expect the DoL to upwardly revise it next week.

However, the AP called this a “sign of a stable job market” on its Twitter feed a few minutes after the release, and Reuters seems a little more excited by it (via Steve Eggleston):

The number of Americans lining up for new jobless benefits fell last week for the first time since April, a reminder that the wounded labor market is still slowly healing.

Initial claims for state unemployment benefits dropped 12,000 to a seasonally adjusted 377,000, the Labor Department said on Thursday. That was spot on the median forecast in a Reuters poll.

The government revised the prior week’s figure up to 389,000 from the previously reported 383,000.

Prior to last week, claims had risen in four consecutive weeks, adding to concerns over several months of lackluster hiring data. While the country emerged from a deep recession three years ago, the jobless rate last month was 8.2 percent, well above its long-term historical average.

Still, most of the recent increases in new jobless claims were marginal and the overall level of claims has held at levels consistent with a modest recovery in the labor market.

Well, there was that 26,000 claim increase at the end of March, the one that pushed the series into the 370K-380K band two months ago.  That was more than twice as much as this week’s drop between an upwardly-revised number and a yet-to-be-revised number.  If that was “marginal,” then today’s change is relatively meaningless.  It’s a continuation of the status quo.  Maybe that’s what the AP means by “stable,” but stability at a 30-year low in the civilian participation ratio and high unemployment isn’t exactly an aspirational goal.

While we’re on the subject of reporting on Department of Labor numbers, many readers may be unaware that the DoL has changed its rules.  From now on, the government agency will forbid media outlets from bringing their own computers to the BLS “lockups” where the data is under embargo, due to concerns over market vulnerability.  Instead, the government will force them to use the BLS’ own computer systems to report on government data, which raises all sorts of questions about government intrusion on free speech and the free press:

According to Carl Fillichio, senior adviser for communications and public affairs at Labor, changing technology and the advent of global instant financial trading prompted the department to conduct its first review of security procedures in more than a decade. “Algorithmic trading introduces new security variables into a lockup system not originally designed to guard against market-moving disruptions that could be caused by the release of government data to certain traders just seconds before the rest of the general public,” he told lawmakers at the hearing. “A few years ago, a few seconds here or there would not have had much of an impact. Today, fractions of a second can equate to millions or even billions of dollars in market movements.”

In April 2011, the department contracted with Sandia National Laboratories to study system vulnerabilities in the BLS lockups. In an August 2011 report, which has not been released to Congress or the public, lab officials recommended a series of changes. Among them, journalists would no longer be permitted to use their own computers, software, or cables — only pens and notebooks. They would rely strictly on a government transmission over the Internet for the data and would no longer be permitted to exit the lockup room before the news embargo lifts.

Media groups have until mid-June to remove their equipment from the lockup room and the first run under the new procedures is scheduled for July 6, when the next national jobless numbers are scheduled to be released.

Members of the media — in the unusual role of appearing at hearings as witnesses — panned the new arrangements.

Daniel Moss, executive director of Bloomberg News, said the department’s presentation of the new system was unprecedented in that it was done without honoring the 1946 Administrative Procedure Act and presented as nonnegotiable. He said using only government Internet feeds “creates a single point of failure” — many news organizations have backup systems in case of malfunction. Moss also complained that “the government would literally own reporters’ notebooks, making them write on government computers, giving the government unfettered access.” This is a First Amendment violation and a risk to national security in the form of cyberattacks, he said. He also criticized the Labor Department for failing to explain the system openly.

That’s okay.  Maybe they’ll upwardly revise the number of computers each week.